Large Language Models and Stock Investing: Is the Human Factor Required?

Paper: arXiv 2603.19944

Authors: Ricardo Crisostomo, Diana Mykhalyuk

Abstract

This paper investigates whether large language models (LLMs) can generate reliable stock market predictions. We evaluate four state-of-the-art models - ChatGPT, Gemini, DeepSeek, and Perplexity - across three prompting strategies: a naive query, a structured approach, and chain-of-thought reasoning. Our results show that LLM-generated recommendations are hindered by recurring reasoning failures, including financial misconceptions, carryover errors, and reliance on outdated or hallucinated information. When appropriately guided and supervised, LLMs demonstrate the capacity to outperform the market, but realizing LLMs’ full potential requires substantial human oversight. We also find that grounding stock recommendations in official regulatory filings increases their forecasting accuracy. Overall, our findings underscore the need for robust safeguards and validation when deploying LLMs in financial markets.

Complexity vs Empirical Score

• Math Complexity: 2.0/10
• Empirical Rigor: 6.0/10 (capped from 6.5: transaction costs not modelled; see penalty flags)
• Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper presents a timely empirical evaluation of multiple modern LLMs across distinct prompting strategies for stock selection. While it lacks formal quantitative or mathematical modeling, it offers strong empirical insights into model failure modes and highlights the critical role of human oversight and RAG-based grounding in SEC filings.

Research Flowchart

  flowchart TD
    A["Research Question: Can LLMs reliably predict stock markets without human intervention?"] --> B["Tested 4 LLMs: ChatGPT, Gemini, DeepSeek, Perplexity"]
    B --> C["Evaluated 3 Prompt Strategies: Naive, Structured, Chain-of-Thought"]
    C --> D["Inputs: Prompting Frameworks & Official Regulatory Filings"]
    D --> E["Analysis of Predictions, Reasoning Failures & Market Performance"]
    E --> F["Findings: Unguided LLMs suffer from hallucinations, misconceptions & errors"]
    E --> G["Outcome: LLMs outperform market only with Human Oversight & Regulatory Grounding"]